summaryrefslogtreecommitdiff
path: root/content/fs_backing_store.c
blob: fde17ed62fc7556a65255ca7b8ce12448aedc780 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
/*
 * Copyright 2014 Vincent Sanders <vince@netsurf-browser.org>
 *
 * This file is part of NetSurf, http://www.netsurf-browser.org/
 *
 * NetSurf is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; version 2 of the License.
 *
 * NetSurf is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */

/** \file
 * Low-level resource cache persistent storage implementation.
 *
 * file based backing store.
 *
 * \todo Consider improving eviction sorting to include objects size
 *         and remaining lifetime and other cost metrics.
 *
 * \todo make backing store have a more efficient small object storage.
 *
 */

#include <unistd.h>
#include <string.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <fcntl.h>
#include <errno.h>
#include <time.h>
#include <stdlib.h>

#include "utils/filepath.h"
#include "utils/file.h"
#include "utils/nsurl.h"
#include "utils/log.h"
#include "utils/utils.h"
#include "utils/messages.h"
#include "desktop/gui.h"

#include "content/backing_store.h"

/** Default number of bits of the ident to use in index hash */
#define DEFAULT_IDENT_SIZE 20

/** Default number of bits to use for an entry index. */
#define DEFAULT_ENTRY_SIZE 16

/** Backing store file format version */
#define CONTROL_VERSION 110

/** Get address from ident */
#define BS_ADDRESS(ident, state) ((ident) & ((1 << state->ident_bits) - 1))

/** Lookup store entry index from ident */
#define BS_ENTRY_INDEX(ident, state) state->addrmap[(ident) & ((1 << state->ident_bits) - 1)]

/** Get store entry from ident. */
#define BS_ENTRY(ident, state) state->entries[state->addrmap[(ident) & ((1 << state->ident_bits) - 1)]]

enum store_entry_flags {
	STORE_ENTRY_FLAG_NONE = 0,
};

/**
 * The type used to store index values refering to store entries. Care
 * must be taken with this type as it is used to build address to
 * entry mapping so changing the size will have large impacts on
 * memory usage.
 */
typedef uint16_t entry_index_t;

/**
 * The type used as a binary identifier for each entry derived from
 * the url. A larger identifier will have fewer collisions but
 * requires proportionately more storage.
 */
typedef uint32_t entry_ident_t;

/**
 * Backing store object index entry.
 *
 * @note Order is important to avoid structure packing overhead.
 */
struct store_entry {
	int64_t last_used; /**< unix time the entry was last used */
	entry_ident_t ident; /**< entry identifier */
	uint32_t data_alloc; /**< currently allocated size of data on disc */
	uint32_t meta_alloc; /**< currently allocated size of metadata on disc */
	uint16_t use_count; /**< number of times this entry has been accessed */
	uint16_t flags; /**< entry flags (unused) */
	uint16_t data_block; /**< small object data block entry (unused) */
	uint16_t meta_block; /**< small object meta block entry (unused) */
};

/**
 * Parameters controlling the backing store.
 */
struct store_state {
	char *path; /**< The path to the backing store */
	size_t limit; /**< The backing store upper bound target size */
	size_t hysteresis; /**< The hysteresis around the target size */

	unsigned int ident_bits; /**< log2 number of bits to use for address. */

	struct store_entry *entries; /**< store entries. */
	unsigned int entry_bits; /**< log2 number of bits in entry index. */
	unsigned int last_entry; /**< index of last usable entry. */

	/** flag indicating if the entries have been made persistant
	 * since they were last changed.
	 */
	bool entries_dirty;

	/** URL identifier to entry index mapping.
	 *
	 * This is an open coded index on the entries url field and
	 * provides a computationaly inexpensive way to go from the
	 * url to an entry.
	 */
	entry_index_t *addrmap;

	uint64_t total_alloc; /**< total size of all allocated storage. */

	size_t hit_count; /**< number of cache hits */
	uint64_t hit_size; /**< size of storage served */
	size_t miss_count; /**< number of cache misses */

};

/**
 * Global storage state.
 *
 * @todo Investigate if there is a way to have a context rather than
 * use a global.
 */
struct store_state *storestate;



/**
 * Remove a backing store entry from the entry table.
 *
 * This finds the store entry associated with the given key and
 * removes it from the table. The removed entry is returned but is
 * only valid until the next set_store_entry call.
 *
 * @param state The store state to use.
 * @param url The value used as the unique key to search entries for.
 * @param bse Pointer used to return value.
 * @return NSERROR_OK and bse updated on succes or NSERROR_NOT_FOUND
 *         if no entry coresponds to the url.
 */
static nserror
remove_store_entry(struct store_state *state,
		   entry_ident_t ident,
		   struct store_entry **bse)
{
	entry_index_t sei; /* store entry index */

	sei = BS_ENTRY_INDEX(ident, state);
	if (sei == 0) {
		LOG(("ident 0x%08x not in index", ident));
		return NSERROR_NOT_FOUND;
	}

	if (state->entries[sei].ident != ident) {
		/* entry ident did not match */
		LOG(("ident 0x%08x did not match entry index %d", ident, sei));
		return NSERROR_NOT_FOUND;
	}

	/* sei is entry to be removed, we swap it to the end of the
	 * table so there are no gaps and the returned entry is held
	 * in storage with reasonable lifetime.
	 */

	/* remove entry from map */
	BS_ENTRY_INDEX(ident, state) = 0;

	/* global allocation accounting  */
	state->total_alloc -= state->entries[sei].data_alloc;
	state->total_alloc -= state->entries[sei].meta_alloc;

	state->last_entry--;

	if (sei == state->last_entry) {
		/* the removed entry was the last one, how convenient */
		*bse = &state->entries[sei];
	} else {
		/* need to swap entries */
		struct store_entry tent;

		tent = state->entries[sei];
		state->entries[sei] = state->entries[state->last_entry];
		state->entries[state->last_entry] = tent;

		/* update map for moved entry */
		BS_ENTRY_INDEX(state->entries[sei].ident, state) = sei;

		*bse = &state->entries[state->last_entry];
	}

	return NSERROR_OK;
}


/**
 * Generate a filename for an object.
 *
 * this generates the filename for an object on disc. It is necessary
 * for this to generate a filename which conforms to the limitations
 * of all the filesystems the cache can be placed upon.
 *
 * From http://en.wikipedia.org/wiki/Comparison_of_file_systems#Limits
 * the relevant subset is:
 *  - path elements no longer than 8 characters
 *  - acceptable characters are A-Z, 0-9
 *  - short total path lengths (255 or less)
 *
 * The short total path lengths mean the encoding must represent as
 * much data as possible in the least number of characters.
 *
 * To achieve all these goals we use RFC4648 base32 encoding which packs
 * 5bits into each character of the filename.
 *
 * @note Version 1.00 of the cache implementation used base64 to
 * encode this, however that did not meet the requirement for only
 * using uppercase characters.
 *
 * @param state The store state to use.
 * @param ident The identifier to use.
 * @return The filename string or NULL on allocation error.
 */
static char *
store_fname(struct store_state *state,
	    entry_ident_t ident,
	    enum backing_store_flags flags)
{
	char *fname = NULL;
	uint8_t b32u_i[8]; /* base32 encoded ident */
	uint8_t b32u_d[6][2]; /* base64 ident as separate components */
	const char *dat;

	/* RFC4648 base32 encoding table */
	static const uint8_t encoding_table[] = {
		'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',
		'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P',
		'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
		'Y', 'Z', '2', '3', '4', '5', '6', '7'
	};

	/* base32 encode ident */
	b32u_i[0] = b32u_d[0][0] = encoding_table[(ident      ) & 0x1f];
	b32u_i[1] = b32u_d[1][0] = encoding_table[(ident >>  5) & 0x1f];
	b32u_i[2] = b32u_d[2][0] = encoding_table[(ident >> 10) & 0x1f];
	b32u_i[3] = b32u_d[3][0] = encoding_table[(ident >> 15) & 0x1f];
	b32u_i[4] = b32u_d[4][0] = encoding_table[(ident >> 20) & 0x1f];
	b32u_i[5] = b32u_d[5][0] = encoding_table[(ident >> 25) & 0x1f];
	b32u_i[6] = encoding_table[(ident >> 30) & 0x1f];
	/* null terminate strings */
	b32u_i[7] = b32u_d[0][1] = b32u_d[1][1] = b32u_d[2][1] =
		b32u_d[3][1] = b32u_d[4][1] = b32u_d[5][1] = 0;

	if ((flags & BACKING_STORE_META) != 0) {
		dat = "m"; /* metadata */
	} else {
		dat = "d"; /* data */
	}

	/* number of chars with usefully encoded data in base 32 */
	switch(((state->ident_bits + 4) / 5)) {
	case 1:
		netsurf_mkpath(&fname, NULL, 3, state->path, dat,
			       b32u_i);
		break;

	case 2:
		netsurf_mkpath(&fname, NULL, 4, state->path, dat,
			       b32u_d[0],
			       b32u_i);
		break;

	case 3:
		netsurf_mkpath(&fname, NULL, 5, state->path, dat,
			       b32u_d[0], b32u_d[1],
			       b32u_i);
		break;

	case 4:
		netsurf_mkpath(&fname, NULL, 6, state->path, dat,
			       b32u_d[0], b32u_d[1], b32u_d[2],
			       b32u_i);
		break;

	case 5:
		netsurf_mkpath(&fname, NULL, 7, state->path, dat,
			       b32u_d[0], b32u_d[1], b32u_d[2], b32u_d[3],
			       b32u_i);
		break;

	case 6:
		netsurf_mkpath(&fname, NULL, 8, state->path, dat,
			       b32u_d[0], b32u_d[1], b32u_d[2], b32u_d[3],
			       b32u_d[4],
			       b32u_i);
		break;

	case 7:
		netsurf_mkpath(&fname, NULL, 9, state->path, dat,
			       b32u_d[0], b32u_d[1], b32u_d[2], b32u_d[3],
			       b32u_d[4], b32u_d[5],
			       b32u_i);
		break;

	default:
		assert("Invalid path depth in store_fname()" == NULL);
	}

	return fname;
}


/**
 * Remove the entry and files associated with an identifier.
 *
 * @param state The store state to use.
 * @param ident The identifier to use.
 * @return NSERROR_OK on sucess or error code on failure.
 */
static nserror
unlink_ident(struct store_state *state, entry_ident_t ident)
{
	char *fname;
	nserror ret;
	struct store_entry *bse;

	/* LOG(("ident %08x", ident)); */

	/* use the url hash as the entry identifier */
	ret = remove_store_entry(state, ident, &bse);
	if (ret != NSERROR_OK) {
		/* LOG(("entry not found")); */
		return ret;
	}

	fname = store_fname(state, bse->ident, BACKING_STORE_META);
	if (fname == NULL) {
		return NSERROR_NOMEM;
	}
	unlink(fname);
	free(fname);

	fname = store_fname(state, bse->ident, BACKING_STORE_NONE);
	if (fname == NULL) {
		return NSERROR_NOMEM;
	}
	unlink(fname);
	free(fname);

	return NSERROR_OK;
}


/**
 * Quick sort comparison.
 */
static int compar(const void *va, const void *vb)
{
	const struct store_entry *a = &BS_ENTRY(*(entry_ident_t *)va, storestate);
	const struct store_entry *b = &BS_ENTRY(*(entry_ident_t *)vb, storestate);

	if (a->use_count < b->use_count) {
		return -1;
	} else if (a->use_count > b->use_count) {
		return 1;
	}
	/* use count is the same - now consider last use time */

	if (a->last_used < b->last_used) {
		return -1;
	} else if (a->last_used > b->last_used) {
		return 1;
	}

	/* they are the same */
	return 0;
}


/**
 * Evict entries from backing store as per configuration.
 *
 * Entries are evicted to ensure the cache remains within the
 * configured limits on size and number of entries.
 *
 * The approach is to check if the cache limits have been exceeded and
 * if so build and sort list of entries to evict. The list is sorted
 * by use count and then by age, so oldest object with least number of uses
 * get evicted first.
 *
 * @param state The store state to use.
 * @return NSERROR_OK on success or error code on failure.
 */
static nserror store_evict(struct store_state *state)
{
	entry_ident_t *elist; /* sorted list of entry identifiers */
	unsigned int ent;
	unsigned int ent_count;
	size_t removed; /* size of removed entries */
	nserror ret = NSERROR_OK;

	/* check if the cache has exceeded configured limit */
	if ((state->total_alloc < state->limit) &&
	    (state->last_entry < (1U << state->entry_bits))) {
		/* cache within limits */
		return NSERROR_OK;
	}

	LOG(("Evicting entries to reduce %d by %d",
	     state->total_alloc, state->hysteresis));

	/* allocate storage for the list */
	elist = malloc(sizeof(entry_ident_t) * state->last_entry);
	if (elist == NULL) {
		return NSERROR_NOMEM;
	}

	/* sort the list avoiding entry 0 which is the empty sentinel */
	for (ent = 1; ent < state->last_entry; ent++) {
		elist[ent - 1] = state->entries[ent].ident;
	}
	ent_count = ent - 1; /* important to keep this as the entry count will change when entries are removed */
	qsort(elist, ent_count, sizeof(entry_ident_t), compar);

	/* evict entries in listed order */
	removed = 0;
	for (ent = 0; ent < ent_count; ent++) {

		removed += BS_ENTRY(elist[ent], state).data_alloc;
		removed += BS_ENTRY(elist[ent], state).meta_alloc;

		ret = unlink_ident(state, elist[ent]);
		if (ret != NSERROR_OK) {
			break;
		}

		if (removed > state->hysteresis) {
			break;
		}
	}

	free(elist);

	LOG(("removed %d in %d entries", removed, ent));

	return ret;
}


/**
 * Lookup a backing store entry in the entry table from a url.
 *
 * This finds the store entry associated with the given
 * key. Additionally if an entry is found it updates the usage data
 * about the entry.
 *
 * @param state The store state to use.
 * @param url The value used as the unique key to search entries for.
 * @param bse Pointer used to return value.
 * @return NSERROR_OK and bse updated on success or NSERROR_NOT_FOUND
 *         if no entry corresponds to the url.
 */
static nserror
get_store_entry(struct store_state *state, nsurl *url, struct store_entry **bse)
{
	entry_ident_t ident;
	unsigned int sei; /* store entry index */

	LOG(("url:%s", nsurl_access(url)));

	/* use the url hash as the entry identifier */
	ident = nsurl_hash(url);

	sei = BS_ENTRY_INDEX(ident, state);

	if (sei == 0) {
		return NSERROR_NOT_FOUND;
	}

	if (state->entries[sei].ident != ident) {
		/* entry ident did not match */
		LOG(("ident did not match entry"));
		return NSERROR_NOT_FOUND;
	}

	*bse = &state->entries[sei];

	state->entries[sei].last_used = time(NULL);
	state->entries[sei].use_count++;

	state->entries_dirty = true;

	return NSERROR_OK;
}


/**
 * Set a backing store entry in the entry table from a url.
 *
 * This creates a backing store entry in the entry table for a url.
 *
 * @param url The value used as the unique key to search entries for.
 * @param bse Pointer used to return value.
 * @return NSERROR_OK and \a bse updated on success or NSERROR_NOT_FOUND
 *         if no entry coresponds to the url.
 */
static nserror
set_store_entry(struct store_state *state,
		nsurl *url,
		enum backing_store_flags flags,
		const uint8_t *data,
		const size_t datalen,
		struct store_entry **bse)
{
	entry_ident_t ident;
	entry_index_t sei; /* store entry index */
	struct store_entry *se;
	nserror ret;
	bool isrep; /* is the store repalcing an existing entry or not */

	LOG(("url:%s", nsurl_access(url)));

	/* evict entries as required and ensure there is at least one
	 * new entry available.
	 */
	ret = store_evict(state);
	if (ret != NSERROR_OK) {
		return ret;
	}

	/* use the url hash as the entry identifier */
	ident = nsurl_hash(url);

	sei = BS_ENTRY_INDEX(ident, state);

	/** @todo Should this deal with cache eviction? */

	if (sei == 0) {
		/* allocating the next available entry */
		sei = state->last_entry;
		state->last_entry++;
		BS_ENTRY_INDEX(ident, state) = sei;
		isrep = false;
	} else {
		/* updating or replacing existing entry */
		/** @todo should we be checking the entry ident
		 * matches the url. Thats a collision in the address
		 * mapping right? and is it important?
		 */
		isrep = true;
	}

	se = &state->entries[sei];

	se->ident = ident;
	se->flags = STORE_ENTRY_FLAG_NONE;
	se->use_count = 1;
	se->last_used = time(NULL);

	/* account for allocation */
	if ((flags & BACKING_STORE_META) != 0) {
		if (isrep) {
			state->total_alloc -= se->meta_alloc;
		} else {
			se->data_alloc = 0;
		}
		se->meta_alloc = datalen;
	} else {
		if (isrep) {
			state->total_alloc -= se->data_alloc;
		} else {
			se->meta_alloc = 0;
		}
		se->data_alloc = datalen;
	}
	state->total_alloc += datalen;

	state->entries_dirty = true;

	*bse = se;

	return NSERROR_OK;
}




/**
 * Open a file using a store ident.
 *
 * @param state The store state to use.
 * @param ident The identifier of the file to open.
 * @param flags The backing store flags.
 * @pram openflags The flags used with the open call.
 * @return An fd from the open call or -1 on error.
 */
static int
store_open(struct store_state *state,
	   uint32_t ident,
	   enum backing_store_flags flags,
	   int openflags)
{
	char *fname;
	nserror ret;
	int fd;

	fname = store_fname(state, ident, flags);
	if (fname == NULL) {
		LOG(("filename error"));
		return -1;
	}

	/** @todo mkdir only on write flag */
	/* ensure path to file is usable */
	ret = netsurf_mkdir_all(fname);
	if (ret != NSERROR_OK) {
		LOG(("file path \"%s\" could not be created", fname));
		free(fname);
		return -1;
	}

	LOG(("opening %s", fname));
	fd = open(fname, openflags, S_IRUSR | S_IWUSR);

	free(fname);

	return fd;
}

/**
 * Construct address ident to filesystem entry map
 *
 * To allow a filesystem entry to be found from it's identifier we
 * construct an mapping index. This is a hash map from the entries URL
 * (its unique key) to filesystem entry.
 *
 * As the entire entry list must be iterated over to construct the map
 * we also compute the total storage in use.
 *
 * @param state The backing store global state.
 * @return NSERROR_OK on sucess or NSERROR_NOMEM if the map storage
 *         could not be allocated.
 */
static nserror
build_entrymap(struct store_state *state)
{
	unsigned int eloop;

	LOG(("Allocating %d bytes for max of %d buckets",
	     (1 << state->ident_bits) * sizeof(entry_index_t),
	     1 << state->ident_bits));

	state->addrmap = calloc(1 << state->ident_bits, sizeof(entry_index_t));
	if (state->addrmap == NULL) {
		return NSERROR_NOMEM;
	}

	state->total_alloc = 0;

	for (eloop = 1; eloop < state->last_entry; eloop++) {
		/*
		LOG(("entry:%d ident:0x%08x used:%d",
		     eloop,
		     BS_ADDRESS(state->entries[eloop].ident, state),
		     state->entries[eloop].use_count));
		*/

		/* update the address map to point at the entry */
		BS_ENTRY_INDEX(state->entries[eloop].ident, state) = eloop;

		/* account for the storage space */
		state->total_alloc += state->entries[eloop].data_alloc +
			state->entries[eloop].meta_alloc;
	}

	return NSERROR_OK;
}

/**
 * Write filesystem entries to file.
 *
 * @todo consider atomic replace using rename.
 *
 * @param state The backing store state to read the entries from.
 * @return NSERROR_OK on sucess or error code on faliure.
 */
static nserror write_entries(struct store_state *state)
{
	int fd;
	char *fname = NULL;
	ssize_t written;
	nserror ret;

	if (state->entries_dirty == false) {
		/* entries have not been updated since last write */
		return NSERROR_OK;
	}

	ret = netsurf_mkpath(&fname, NULL, 2, state->path, "entries");
	if (ret != NSERROR_OK) {
		return ret;
	}

	fd = open(fname, O_RDWR | O_CREAT | O_TRUNC, S_IRUSR | S_IWUSR);
	free(fname);
	if (fd == -1) {
		return NSERROR_SAVE_FAILED;
	}

	written = write(fd, state->entries,
			state->last_entry * sizeof(struct store_entry));
	close(fd);
	if (written < 0) {
		/** @todo Delete the file? */
		return NSERROR_SAVE_FAILED;
	}

	return NSERROR_OK;
}

/**
 * Read description entries into memory.
 *
 * @param state The backing store state to put the loaded entries in.
 * @return NSERROR_OK on sucess or error code on faliure.
 */
static nserror
read_entries(struct store_state *state)
{
	int fd;
	ssize_t rd;
	size_t entries_size;
	char *fname = NULL;
	nserror ret;

	ret = netsurf_mkpath(&fname, NULL, 2, state->path, "entries");
	if (ret != NSERROR_OK) {
		return ret;
	}

	entries_size = (1 << state->entry_bits) * sizeof(struct store_entry);

	LOG(("Allocating %d bytes for max of %d entries",
	     entries_size, 1 << state->entry_bits));

	state->entries = calloc(1, entries_size);
	if (state->entries == NULL) {
		free(fname);
		return NSERROR_NOMEM;
	}

	fd = open(fname, O_RDWR);
	free(fname);
	if (fd != -1) {
		rd = read(fd, state->entries, entries_size);
		close(fd);
		if (rd > 0) {
			state->last_entry = rd / sizeof(struct store_entry);
			LOG(("Read %d entries", state->last_entry));
		}
	} else {
		/* could rebuild entries from fs */
		state->last_entry = 1;
	}
	return NSERROR_OK;
}

/**
 * Write the cache tag file.
 *
 * @param state The cache state.
 * @return NSERROR_OK on sucess or error code on faliure.
 */
static nserror
write_cache_tag(struct store_state *state)
{
	FILE *fcachetag;
	nserror ret;
	char *fname = NULL;

	ret = netsurf_mkpath(&fname, NULL, 2, state->path, "CACHEDIR.TAG");
	if (ret != NSERROR_OK) {
		return ret;
	}

	fcachetag = fopen(fname, "wb");

	free(fname);

	if (fcachetag == NULL) {
		return NSERROR_NOT_FOUND;
	}

	fprintf(fcachetag,
		"Signature: 8a477f597d28d172789f06886806bc55\n"
		"# This file is a cache directory tag created by NetSurf.\n"
		"# For information about cache directory tags, see:\n"
		"#	http://www.brynosaurus.com/cachedir/\n");

	fclose(fcachetag);

	return NSERROR_OK;
}

/**
 * Write the control file for the current state.
 *
 * @param state The state to write to the control file.
 * @return NSERROR_OK on sucess or error code on faliure.
 */
static nserror
write_control(struct store_state *state)
{
	FILE *fcontrol;
	nserror ret;
	char *fname = NULL;

	ret = netsurf_mkpath(&fname, NULL, 2, state->path, "control");
	if (ret != NSERROR_OK) {
		return ret;
	}

	LOG(("writing control file \"%s\"", fname));

	ret = netsurf_mkdir_all(fname);
	if (ret != NSERROR_OK) {
		free(fname);
		return ret;
	}

	fcontrol = fopen(fname, "wb");

	free(fname);

	if (fcontrol == NULL) {
		return NSERROR_NOT_FOUND;
	}

	fprintf(fcontrol, "%u%c", CONTROL_VERSION, 0);
	fprintf(fcontrol, "%u%c", state->entry_bits, 0);
	fprintf(fcontrol, "%u%c", state->ident_bits, 0);
	fprintf(fcontrol, "%u%c", state->last_entry, 0);

	fclose(fcontrol);

	return NSERROR_OK;
}


/**
 * Read and parse the control file.
 *
 * @param state The state to read from the control file.
 * @return NSERROR_OK on sucess or error code on faliure.
 */
static nserror
read_control(struct store_state *state)
{
	nserror ret;
	FILE *fcontrol;
	unsigned int ctrlversion;
	unsigned int addrbits;
	unsigned int entrybits;
	char *fname = NULL;

	ret = netsurf_mkpath(&fname, NULL, 2, state->path, "control");
	if (ret != NSERROR_OK) {
		return ret;
	}

	LOG(("opening control file \"%s\"", fname));

	fcontrol = fopen(fname, "rb");

	free(fname);

	if (fcontrol == NULL) {
		/* unable to open control file */
		if (errno == ENOENT) {
			return NSERROR_NOT_FOUND;
		} else {
			return NSERROR_INIT_FAILED;
		}
	}

	/* read control and setup new state */

	/* first line is version */
	if (fscanf(fcontrol, "%u", &ctrlversion) != 1) {
		goto control_error;
	}

	if (ctrlversion != CONTROL_VERSION) {
		goto control_error;
	}

	if (fgetc(fcontrol) != 0) {
		goto control_error;
	}

	/* second line is log2 max number of entries */
	if (fscanf(fcontrol, "%u", &entrybits) != 1) {
		goto control_error;
	}
	if (fgetc(fcontrol) != 0) {
		goto control_error;
	}

	/* second line is log2 size of address hash */
	if (fscanf(fcontrol, "%u", &addrbits) != 1) {
		goto control_error;
	}
	if (fgetc(fcontrol) != 0) {
		goto control_error;
	}

	fclose(fcontrol);

	state->entry_bits = entrybits;
	state->ident_bits = addrbits;

	return NSERROR_OK;

control_error: /* problem with the control file */

	fclose(fcontrol);

	return NSERROR_INIT_FAILED;
}




/* Functions exported in the backing store table */

/**
 * Initialise the backing store.
 *
 * @param parameters to configure backing store.
 * @return NSERROR_OK on success or error code on faliure.
 */
static nserror
initialise(const struct llcache_store_parameters *parameters)
{
	struct store_state *newstate;
	nserror ret;

	/* check backing store is not already initialised */
	if (storestate != NULL) {
		return NSERROR_INIT_FAILED;
	}

	/* if we are not allowed any space simply give up on init */
	if (parameters->limit == 0) {
		return NSERROR_OK;
	}

	/* if the path to the cache directory is not set do not init */
	if (parameters->path == NULL) {
		return NSERROR_OK;
	}

	/* allocate new store state and set defaults */
	newstate = calloc(1, sizeof(struct store_state));
	if (newstate == NULL) {
		return NSERROR_NOMEM;
	}

	newstate->path = strdup(parameters->path);
	newstate->limit = parameters->limit;
	newstate->hysteresis = parameters->hysteresis;

	if (parameters->address_size == 0) {
		newstate->ident_bits = DEFAULT_IDENT_SIZE;
	} else {
		newstate->ident_bits = parameters->address_size;
	}

	if (parameters->entry_size == 0) {
		newstate->entry_bits = DEFAULT_ENTRY_SIZE;
	} else {
		newstate->entry_bits = parameters->entry_size;
	}

	ret = read_control(newstate);
	if (ret != NSERROR_OK) {
		LOG(("read control failed %s", messages_get_errorcode(ret)));
		ret = write_control(newstate);
		if (ret == NSERROR_OK) {
			write_cache_tag(newstate);
		}
	}
	if (ret != NSERROR_OK) {
		/* that went well obviously */
		free(newstate->path);
		free(newstate);
		return ret;
	}

	/* ensure the maximum number of entries can be represented in
	 * the type available to store it.
	 */
	if (newstate->entry_bits > (8 * sizeof(entry_index_t))) {
		newstate->entry_bits = (8 * sizeof(entry_index_t));
	}

	/* read filesystem entries */
	ret = read_entries(newstate);
	if (ret != NSERROR_OK) {
		/* that went well obviously */
		free(newstate->path);
		free(newstate);
		return ret;
	}

	/* build entry hash map */
	ret = build_entrymap(newstate);
	if (ret != NSERROR_OK) {
		/* that obviously went well  */
		free(newstate->path);
		free(newstate);
		return ret;
	}

	storestate = newstate;

	LOG(("FS backing store init successful"));

	LOG(("path:%s limit:%d hyst:%d addr:%d entries:%d", newstate->path, newstate->limit, newstate->hysteresis, newstate->ident_bits, newstate->entry_bits));
	LOG(("Using %d/%d", newstate->total_alloc, newstate->limit));

	return NSERROR_OK;
}


/**
 * Finalise the backing store.
 *
 * @return NSERROR_OK on success.
 */
static nserror
finalise(void)
{
	if (storestate != NULL) {
		write_entries(storestate);

		/* avoid division by zero */
		if (storestate->miss_count == 0) {
			storestate->miss_count = 1;
		}
		LOG(("hits:%d misses:%d hit ratio:%d returned:%d bytes",
		     storestate->hit_count, storestate->miss_count,
		     storestate->hit_count / storestate->miss_count,
		     storestate->hit_size));

		free(storestate->path);
		free(storestate);
		storestate = NULL;
	}
	return NSERROR_OK;
}


/**
 * Place an object in the backing store.
 *
 * @param url The url is used as the unique primary key for the data.
 * @param flags The flags to control how the object is stored.
 * @param data The objects source data.
 * @param datalen The length of the \a data.
 * @return NSERROR_OK on success or error code on faliure.
 */
static nserror
store(nsurl *url,
      enum backing_store_flags flags,
      const uint8_t *data,
      const size_t datalen)
{
	nserror ret;
	struct store_entry *bse;
	ssize_t written;
	int fd;

	/* check backing store is initialised */
	if (storestate == NULL) {
		return NSERROR_INIT_FAILED;
	}

	/* set the store entry up */
	ret = set_store_entry(storestate, url, flags, data, datalen, &bse);
	if (ret != NSERROR_OK) {
		LOG(("store entry setting failed"));
		return ret;
	}

	fd = store_open(storestate, bse->ident, flags, O_CREAT | O_WRONLY);
	if (fd < 0) {
		perror("");
		LOG(("Open failed %d",fd));
		return NSERROR_SAVE_FAILED;
	}


	LOG(("Writing %d bytes from %p", datalen, data));
	written = write(fd, data, datalen);

	close(fd);
	if (written < 0 || (size_t) written < datalen) {
		/** @todo Delete the file? */
		return NSERROR_SAVE_FAILED;
	}

	return NSERROR_OK;
}

/**
 * Retrive an object from the backing store.
 *
 * @param url The url is used as the unique primary key for the data.
 * @param flags The flags to control how the object is stored.
 * @param data The objects data.
 * @param datalen The length of the \a data retrieved.
 * @return NSERROR_OK on success or error code on faliure.
 */
static nserror
fetch(nsurl *url,
      enum backing_store_flags *flags,
      uint8_t **data_out,
      size_t *datalen_out)
{
	nserror ret;
	struct store_entry *bse;
	uint8_t *data;
	size_t datalen;
	int fd;
	ssize_t rd;

	/* check backing store is initialised */
	if (storestate == NULL) {
		return NSERROR_INIT_FAILED;
	}

	ret = get_store_entry(storestate, url, &bse);
	if (ret != NSERROR_OK) {
		LOG(("entry not found"));
		storestate->miss_count++;
		return ret;
	}
	storestate->hit_count++;

	LOG(("retriving cache file for url:%s", nsurl_access(url)));

	fd = store_open(storestate, bse->ident, *flags, O_RDONLY);
	if (fd < 0) {
		LOG(("Open failed"));
		/** @todo should this invalidate the entry? */
		return NSERROR_NOT_FOUND;
	}

	data = *data_out;
	datalen = *datalen_out;

	/* need to deal with buffers */
	if (data == NULL) {
		if (datalen == 0) {
			/* caller did not know the files length */
			if (((*flags) & BACKING_STORE_META) != 0) {
				datalen = bse->meta_alloc;
			} else {
				datalen = bse->data_alloc;
			}
		}

		data = malloc(datalen);
		if (data == NULL) {
			close(fd);
			return NSERROR_NOMEM;
		}
	}

	/** @todo should this check datalen is sufficient */

	LOG(("Reading %d bytes into %p from file", datalen, data));

	/** @todo this read should be an a loop */
	rd = read(fd, data, datalen);
	if (rd <= 0) {
		LOG(("read returned %d", rd));
		close(fd);
		if ((*data_out) == NULL) {
			free(data);
		}
		return NSERROR_NOT_FOUND;
	}

	close(fd);

	storestate->hit_size += datalen;

	*data_out = data;
	*datalen_out = datalen;

	return NSERROR_OK;
}


/**
 * Invalidate a source object from the backing store.
 *
 * The entry (if present in the backing store) must no longer
 * be returned as a result to the fetch or meta operations.
 *
 * @param url The url is used as the unique primary key to invalidate.
 * @return NSERROR_OK on success or error code on faliure.
 */
static nserror
invalidate(nsurl *url)
{
	/* check backing store is initialised */
	if (storestate == NULL) {
		return NSERROR_INIT_FAILED;
	}

	LOG(("url:%s", nsurl_access(url)));

	return unlink_ident(storestate, nsurl_hash(url));
}


static struct gui_llcache_table llcache_table = {
	.initialise = initialise,
	.finalise = finalise,
	.store = store,
	.fetch = fetch,
	.invalidate = invalidate,
};

struct gui_llcache_table *filesystem_llcache_table = &llcache_table;